The most popular and comprehensive Open Source ECM platform
Let us know more
Artificial Intelligence 2.0: Five New Techniques Helping AI to Evolve
Forrester Predictions 2021 said that “the ‘great lockdown’ of 2020 will make the drive for automation in 2021 both inevitable and irreversible. Remote work, new digital muscles, and pandemic constraints will create millions of pragmatic automations in 2021; document extraction, RPA (robotic process automation) from anywhere, drones, and various employee robots will proliferate; and, as expected, the mad dash to automate will bring trouble.”
Forrester sees Artificial Intelligence as entering a new phase, something they are calling AI 2.0. Forrester identifies five areas and new algorithmic techniques that are driving the transformation of AI:
Transformer Network – A kind of deep learning model introduced in 2017 that is used primarily for Natural Language Processing (NLP). Transformer networks are used especially for translation and text summarization.
Synthetic Data to Extend Training Data – Massive amounts of collected data are known to improve AI models but may be hard to come by. Synthetic data can augment existing data and lead to more accurate AI models.
Reinforcement Learning – A type of AI that doesn’t require labeled input data but tries to maximize a computed reward based on possible actions that can be taken.
Federated Learning – A method for combining AI results from existing AI models. It allows new models to be built based on the existing results and enables the sharing and pooling of data from many decentralized locations. It is useful to avoid privacy and bandwidth issues.
Causal Reasoning – AI algorithms are being trained to employ causal reasoning to draw conclusions.
